Hackinger et al. Translational Psychiatry (2018) 8:252 DOI 10.1038/s41398-018-0304-6 Translational Psychiatry ARTICLE Open Access Evidence for genetic contribution to the increased risk of type 2 diabetes in schizophrenia Sophie Hackinger1, Bram Prins 2, Vasiliki Mamakou 3,4, Eleni Zengini4,5, Eirini Marouli6,LukaBrčić7, Ioannis Serafetinidis8, Klea Lamnissou9, Vassilis Kontaxakis10, George Dedoussis11, Fragiskos Gonidakis12, Anastasia Thanopoulou13, Nikolaos Tentolouris14,AspasiaTsezou15 and Eleftheria Zeggini 1,16 Abstract The epidemiologic link between schizophrenia (SCZ) and type 2 diabetes (T2D) remains poorly understood. Here, we investigate the presence and extent of a shared genetic background between SCZ and T2D using genome-wide approaches. We performed a genome-wide association study (GWAS) and polygenic risk score analysis in a Greek sample collection (GOMAP) comprising three patient groups: SCZ only (n = 924), T2D only (n = 822), comorbid SCZ and T2D (n = 505); samples from two separate Greek cohorts were used as population-based controls (n = 1,125). We used genome-wide summary statistics from two large-scale GWAS of SCZ and T2D from the PGC and DIAGRAM consortia, respectively, to perform genetic overlap analyses, including a regional colocalisation test. We show for the first time that patients with comorbid SCZ and T2D have a higher genetic predisposition to both disorders compared to controls. We identify five genomic regions with evidence of colocalising SCZ and T2D signals, three of which contain known loci for both diseases. We also observe a significant excess of shared association signals between SCZ 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; and T2D at nine out of ten investigated p value thresholds. Finally, we identify 29 genes associated with both T2D and SCZ, several of which have been implicated in biological processes relevant to these disorders. Together our results demonstrate that the observed comorbidity between SCZ and T2D is at least in part due to shared genetic mechanisms. Introduction population4—both risk factors for T2D. Antipsychotic Schizophrenia (SCZ) patients are 1.5–2 times more drugs, particularly second generation antipsychotics, are likely to develop type 2 diabetes (T2D) compared to the known to cause metabolic side effects and often lead to general population1. Several explanations for this epide- significant weight gain5. Several studies have found an miologic link have been proposed, including environ- association between psychotropic medication and T2D – mental factors, the use of antipsychotic medication, and/ risk6 8, but it is still unclear to what extent interactions – or shared genetic aetiology1 4. For example, patients with between different medications, life-style and inter-patient severe mental illness often lead a more sedentary life and variability affect this association4. It is conceivable that the are more likely to smoke compared to the general metabolic effects of antipsychotics are partly mediated by genetic predisposition. So far, studies on the genetics of antipsychotic response have been small (n < 400) and 9,10 Correspondence: Eleftheria Zeggini (eleftheria.zeggini@helmholtz-muenchen. unable to identify replicating associations . de) 1Human Genetics, Wellcome Trust Sanger Institute, Hinxton CB10 1HH, UK 2Strangeways Research Laboratory, University of Cambridge, 2 Worts’ Causeway, Cambridge CB1 8RN, UK Full list of author information is available at the end of the article. © The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a linktotheCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Hackinger et al. Translational Psychiatry (2018) 8:252 Page 2 of 10 Table 1 Sample numbers in the three phenotype groups Methods in GOMAP before and after QC Data sets The GOMAP (Genetic Overlap between Metabolic and Sample group Pre-QC Post-QC Psychiatric disorders) study comprises a collection of SCZ 977 924 2,747 DNA samples from four different patient categories: T2D patients, SCZ patients, individuals with both SCZ T2D 885 822 and T2D (referred to from here on as SCZplusT2D), and SCZplusT2D 542 505 individuals with a different psychiatric diagnosis (this last Other 343 331 group is not used in further analyses reported here) (Table Total 2747 2582 1). SCZ patients with and without T2D were recruited at the Dromokaitio Psychiatric Hospital and Dafni Psychia- tric Hospital in Athens. SCZ diagnosis was determined by structured clinical interview of the Diagnostic and Sta- In addition, there is evidence that the increased pre- tistical Manual of Mental Disorders 4th edition (DSM- valence of T2D in patients with SCZ is not purely medi- IV)23. T2D participants were recruited from diabetes cation induced: Proteomic studies have revealed outpatient clinics at Hippokrateio General Hospital and perturbed expression of genes involved in glucose meta- Laiko General Hospital. T2D status was assessed in all bolism in brain tissue and elevated insulin levels in per- participants based on criteria outlined by the American ipheral blood of first-episode SCZ patients compared to Diabetes Association24. All participants gave written controls11,12. More recently, a large study following over informed consent. 2.5 million Danish individuals found that antipsychotic- naïve SCZ patients were three times more likely to Quality control develop T2D than the general population, with anti- A total of 2,474 samples and 538,448 markers were psychotic drug use further increasing that risk13. This, successfully genotyped on the Illumina HumanCoreEx- along with findings from a systematic review and meta- ome 12v1.0 BeadChip (Illumina, San Diego, CA, USA) at analysis14, suggests that impaired glucose homeostasis the Wellcome Trust Sanger Institute, Hinxton, UK. may already be present in drug-naïve SCZ patients. Quality control (QC) of genotype data was performed It is also plausible that the observed overlap between following a standard protocol25 using the PLINK26 soft- SCZ and T2D is due to common susceptibility variants2. ware package. Individuals were removed if they had a call Both diseases are highly polygenic, and genome-wide rate below 90%, discordant values for genotyped and association studies (GWAS) to date have successfully reported sex or had heterozygosity rates deviating more – identified a substantial number of risk loci for T2D15 18 than three standard deviations from the mean. For – and SCZ19 21. Functional analyses showed that risk var- duplicates and related sample pairs (pi_hat > 0.2), we iants for SCZ are enriched for enhancers mapping to excluded one and retained the other. pancreatic beta cells19, and that variants associated with In order to identify potential ethnic outliers, we per- BMI – a key risk factor for T2D – predominately map to formed multidimensional scaling (MDS) on a merged central nervous system pathways22. Genetic research into dataset comprising GOMAP and three other Greek the shared pathobiology of SCZ and T2D has been limited sample collections: TEENAGE27, a collection of adoles- to date, and has mainly focused on patients with one of cents from the general Greek population, HELIC- the two disorders2. If SCZ without T2D comorbidity and POMAK25 and HELIC-MANOLIS28, two Greek isolated SCZ with T2D are partly underpinned by different genetic population cohorts. We removed seven individuals from aetiologies, such study designs will fail to identify risk GOMAP as outliers based on the first and second MDS factors predisposing to the latter. components (Supplementary Figure 1). Here, we investigate the presence of shared genetic risk A total of 2,582 samples passed QC (Supplementary factors for T2D and SCZ using genotype data from a Table 1; samples size of each diagnostic category: SCZ, n novel cohort comprising three patient groups (T2D only, = 924; T2D, n = 822; T2D/SCZ, n = 505; other diagnosis, SCZ only, and comorbid SCZ and T2D), as well as sum- n = 331). mary data from large-scale disease-specific GWAS. First, After removal of individuals failing QC, variants were we conduct genome-wide comparisons between all three filtered for call rates lower than 98%, a Hardy-Weinberg – patient groups, as well as population controls; next, we Equilibrium deviation p value < 1 × 10 4 and cluster assess the genetic overlap between the two disorders using separation scores below 0.4. In addition, we removed X- polygenic risk scores; finally, we use summary statistics chromosomal markers not within the pseudo-autosomal from published GWAS to search for genetic risk factors region with heterozygous haploid genotypes in males. A shared between SCZ and T2D. Hackinger et al. Translational Psychiatry (2018) 8:252 Page 3 of 10 total of 524,271 autosomal and X-chromosomal markers European descent. Three of the 76 variants were